Gartner projects that by 2028 one in four candidate profiles globally could be fake. Whether the eventual figure lands at twenty-five per cent or half that, the direction is not in dispute, and the response does not depend on the number being right. It depends on shifting verification effort from the end of the process to the beginning.
What follows is a staged plan rather than a warning. Most of it costs process discipline rather than money.
What the Projection Rests On
Three forces, all already visible. Generating a plausible application costs almost nothing. Remote hiring removed the incidental verification that a physical process provided. And application volume has risen to a level where genuine scrutiny of every submission is not achievable by reading.
The supporting evidence is uncomfortable rather than speculative. A Gartner survey of around three thousand candidates found a small but non-trivial proportion admitting some form of interview fraud. Experian listed deepfake job candidates among its 2026 fraud threats. Surveys of hiring managers in the United States have found a meaningful minority reporting they have already encountered manipulated video interviews.
Pair that with volume. UK graduate vacancies now attract roughly 140 applications each against about 38 two decades ago, and two thirds of UK hiring managers surveyed by Robert Half said AI-generated applications had lengthened their time to hire. Fraud does not need to be common to be costly when the denominator is that large.
Stage One: Establish Your Actual Exposure
Do not begin with tooling. Begin with an honest map of where you would fail.
- Which roles are fully remote from application to onboarding, with no point at which a person is physically present or verified independently?
- Which roles carry access to systems, funds or personal data on day one?
- Where do you accept referees nominated by the candidate with no independent corroboration?
- Where is identity verified once, early, and never revisited before a start date weeks later?
- Which engagement types sit outside your main process, such as contractors, temporary supply or platform-sourced work?
The highest-risk combination is a remote technical or finance role, sourced at volume, verified once, with privileged access from the first morning. If you have those, start there rather than everywhere.
Stage Two: Make the Sift Evidence-Based
This is the stage with the largest return, and it is a change of method rather than a purchase.
For each essential criterion, classify what the application actually does: evidences the requirement, claims it without evidence, or does not address it. That single distinction changes the economics of fabrication, because inventing a claim is trivial and inventing corroborable specifics is not.
It also produces something useful downstream. A structured sift hands the interviewer a list of what was claimed but unevidenced, which is precisely where probing should concentrate. Most processes generate that information and then throw it away.
Stage Three: Corroborate Where a Register Exists
External registers are the cheapest verification available and the hardest to defeat. Where a role depends on registration or chartered status, check the register rather than the certificate.
Nursing and midwifery, medical and allied health professions, solicitors, and the engineering institutions all maintain searchable registers. A register check takes minutes, cannot be socially engineered, and answers definitively. A PDF certificate answers nothing.
Stage Four: Fix the Reference Process
Candidate-nominated references verified by email are the weakest control in most processes, and the easiest to fabricate convincingly.
Three changes make a material difference. Contact the organisation through a number or domain you established independently, not one supplied in the application. Verify the referee exists in the role claimed before weighing what they say. And ask factual questions with checkable answers, dates, job title, reporting line, rather than inviting an opinion.
Stage Five: Verify Again Before the Start Date
Proxy arrangements depend on nobody checking whether the person who starts is the person who was assessed. Where the gap between assessment and start is weeks, and everything is remote, re-verify identity at onboarding as a matter of routine.
Equipment despatch addresses, repeated deferrals of a start date, and a reluctance to complete onboarding verification are all worth treating as signals rather than administrative friction.
Stage Six: Write It Down and Apply It Uniformly
An anti-fraud process applied selectively becomes an Equality Act 2010 problem. Escalating verification because of a name, an accent or an assumed nationality is unlawful, and it is also ineffective, because it concentrates scrutiny on the wrong variable.
Document the stack, apply every layer to every candidate at the same depth, and record that you did. That record is what makes a rejection defensible and what demonstrates, if challenged, that your process was consistent rather than intuitive.
What Not to Buy
Two categories deserve scepticism. AI writing detectors answer the wrong question and produce false positives that fall hardest on non-native English writers; whether an application was drafted with assistance says nothing about whether the experience behind it is real. Deepfake detection has a place but is an arms race with contested accuracy, and building a process on a control that degrades as generation improves is a poor foundation.
Evidence verification does not degrade. A candidate either can describe the specifics of work they did, or cannot.
Frequently Asked Questions
Is This Really a UK Problem?
Yes, and not only for remote roles. An NHS board publicly reported receiving high volumes of applications from overseas that did not meet eligibility criteria and were thought to be AI-generated. Volume-driven inauthenticity is already arriving in UK public sector recruitment.
Where Is the Highest Risk Concentrated?
Fully remote roles with privileged access, sourced at volume, verified once. Technical, finance and data roles combine all three more often than most.
Does This Mean Trusting Candidates Less?
It means verifying claims rather than reading them. Honest candidates are advantaged by an evidence-based process, because specifics are exactly what they have and fabricators do not.
How Do We Avoid Rejecting Good Candidates?
Assess evidence of capability, not polish of language. Penalising imperfect writing rejects competent people and catches no fraud, since fabricated applications tend to read better than genuine ones.
What Should We Do First?
Make the sift evidence-based and fix the reference process. Those two changes cost no money, address the most common fraud patterns, and improve hiring quality regardless of whether the projection proves accurate.
The Honest Summary
A projection about 2028 is not a reason to act. The reason to act is that the controls which defend against synthetic applicants are the same controls that produce better shortlists from real ones: specific evidence, external corroboration, independent references and consistent application.
Organisations that build that now get the hiring quality benefit immediately and the fraud resilience as a by-product. Organisations that wait will be buying detection tools to compensate for a sift that was never asking the right question.
This article is general information, not legal advice. Survey and forecast figures are attributed to their publishers and should be checked against current releases.
Sources
The Stationery Office. Equality Act 2010.
https://www.legislation.gov.uk/ukpga/2010/15/contents
Home Office. Draft Code of Practice for Employers: Avoiding Unlawful Discrimination While Preventing Illegal Working.
https://www.gov.uk/government/consultations/right-to-work-draft-code-of-practice-for-employers-avoiding-unlawful-discrimination/draft-code-of-practice-for-employers-avoiding-unlawful-discrimination-while-preventing-illegal-working
Nursing and Midwifery Council. Check the Register.
https://www.nmc.org.uk/registration/search-the-register/
Health and Care Professions Council. Check the Register.
https://www.hcpc-uk.org/check-the-register/
Solicitors Regulation Authority. Solicitors Register.
https://www.sra.org.uk/consumers/register/
Information Commissioner's Office. Employment Practices: Recruitment and Selection.
https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/employment/recruitment-and-selection/
The Stationery Office. Border Security, Asylum and Immigration Act 2025, Section 48.
https://www.legislation.gov.uk/ukpga/2025/31/section/48
InsightCircle
